Research map / Graph learning / Knowledge and temporal graphs
Semantic Object Parsing with Graph LSTM
Xiaodan Liang, Xiaohui Shen, Jiashi Feng, Liang Lin, Shuicheng Yan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use adaptive Graph LSTM instead of fixed factorized LG-LSTM | semantic parsing network · PASCAL-Person-Part | average IoU (%): 57.97 -> 60.16 | Improved |
| use confidence-driven node updating instead of BFS (location) (ablation) | Graph LSTM · PASCAL-Person-Part | average IoU (%): 58.63 -> 60.16 | Improved |
| use confidence-driven node updating instead of DFS (confidence) (ablation) | Graph LSTM · PASCAL-Person-Part | average IoU (%): 59.21 -> 60.16 | Improved |
| add adaptive forget gates instead of identical forget gates (ablation) | Graph LSTM · PASCAL-Person-Part | average IoU (%): 59.21 -> 60.16 | Improved |
| add residual connections between Graph LSTM layers (ablation) | Graph LSTM · PASCAL-Person-Part | average IoU (%): 59.12 -> 60.16 | Improved |
| add Graph LSTM instead of superpixel smoothing post-processing on LG-LSTM (ablation) | LG-LSTM · PASCAL-Person-Part | average IoU (%): 58.58 -> 60.16 | Improved |
| add superpixel smoothing post-processing (ablation) | LG-LSTM · PASCAL-Person-Part | average IoU (%): 57.97 -> 58.58 | Improved |
| use confidences of head or torso label to order node updates instead of all foreground labels (ablation) | Graph LSTM · PASCAL-Person-Part | average IoU (%): 60.16 -> 61.03 | Mixed |
| increase superpixels beyond 1,000 per image (ablation) | Graph LSTM · PASCAL-Person-Part | average IoU (%) | No clear effect |
| add 10,000 extra training images | Graph LSTM · ATR | average F-1 score (%): 83.76 -> 88.20 | Improved |
| add 10,000 extra training images | Graph LSTM · Fashionista | average F-1 score (%): 87.57 -> 89.75 | Improved |
| use Graph LSTM instead of LG-LSTM (more) | Graph LSTM (more) · ATR | average F-1 score (%): 84.12 -> 88.20 | Improved |
Closest papers by meaning
- Semantic Object Parsing With Local-Global Long Short-Term Memory (CVPR 2016 · similarity 0.92)
- Interpretable Structure-Evolving LSTM (CVPR 2017 · similarity 0.89)
- Weakly Supervised Visual Semantic Parsing (CVPR 2020 · similarity 0.89)
- Graphonomy: Universal Image Parsing via Graph Reasoning and Transfer (TPAMI 2020 · similarity 0.88)
- SCENIR: Visual Semantic Clarity through Unsupervised Scene Graph Retrieval (ICML 2025 · similarity 0.86)
- Tracing Hyperparameter Dependencies for Model Parsing via Learnable Graph Pooling Network (NeurIPS 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
